查看文章

The recent advent of computational intelligence and the field of deep learning has shown a significant application for the task of efficient navigation of automated and unmanned vehicles. The notion of a robot intelligently deciding a path based on minimalistic spatial knowledge and operating in a collision-free manner illustrates significant real-world importance. This paper offers a novel study for determining the best strategy for robust path planning in a simulated environment sufficed with static obstacles. The paradigm of behavior cloning and imitation learning is extensively explored, these techniques have depicted a better analogy to the human brain hence, justifying the experiments. This paper also conducts extensive tests on the existing technologies as baselines for an unbiased comparison, these algorithms include Rapidly- exploring Random Tree (RRT), A* search algorithm, and an improved rendition of …